OpenAI, Anthropic to Brief UN Security Council on AI Risks

OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei are expected to brief the United Nations Security Council on Sept. 23 as governments examine how increasingly capable AI systems could affect international security.

France, which holds the Security Council presidency this month, is convening the high-level meeting under the “maintenance of international peace and security” agenda. Its concept note focuses on malicious AI use, model misalignment and the risk of losing control over highly capable systems.

Altman, Amodei and Bengio Expected to Brief Council 

French Foreign Minister Jean-Noël Barrot is expected to chair the meeting during the UN General Assembly’s high-level week in New York.

Anticipated briefers include Altman, Amodei, AI researcher Yoshua Bengio and Hugging Face CEO Clément Delangue. Chinese AI developers have also been invited to make statements, although their participation plans could still change. 

The Security Council first held a formal meeting dedicated to AI in July 2023. It returned to the subject in December 2024 and held another high-level AI briefing in September 2025.

OpenAI and Anthropic Incidents Add Cyber-Risk Context 

The meeting follows recent disclosures showing advanced models reaching systems outside their intended testing environments.

OpenAI said in August that models operating during internal cybersecurity evaluations bypassed isolation controls, reached the internet and compromised parts of OpenAI’s research infrastructure and Hugging Face systems. 

The company subsequently paused reinforcement-learning work on some frontier models while strengthening security, monitoring, and alignment controls.

Anthropic separately disclosed four incidents in which Claude models gained unauthorized access to real third-party systems during cybersecurity evaluations. Anthropic said the evaluation environments had been mistakenly connected to the open internet and the models were running without the cyber safeguards used in released products.

Its September threat-intelligence report separately documented suspected state-sponsored groups, criminals and other actors using Claude in cyber, surveillance and influence operations.

Altman Expected to Call for Shared International AI Safety Standards 

OpenAI said Altman’s remarks are expected to focus on international coordination, shared safety standards and measures intended to ensure AI benefits people globally.

Amodei has separately called for AI developers to moderate the pace of capability improvements to give safety measures more time to keep pace.

The Sept. 23 meeting will not create binding global AI rules by itself. It places leading AI developers directly before the Security Council as members consider how cyber threats, malicious use and increasingly autonomous systems fit into international security policy.